Mi Wang
RESEARCH ON MULTI-RESOLUTION SEAMLESS IMAGE DATABASE
Mi WANG, Jianya GONG, Deren LI
National Laboratory for Information Engineering in Surveying, Mapping and Remote Sensing
Wuhan Technical University of Surveying and MappingˈP.R. China
wangmi@rcgis.wtusm.edu.cn,dli@wtusm.edu.cn,jgong@rcgis.wtusm.edu.cn
Working Group IV/1
KEY WORDS: Multi-scale Database, Spatial Databases, Data Structures, Design, Software
ABSTRACT
This paper presents the basic concepts and principles, data structure and high efficient spatial index for MultiResolution Image Database. The database is characterized with arrangement of multi-resource image data and seamless
mosaic, distribution-based storage and management, integration with other spatial database software such as GeoStar
and GeoGrid developed by Wuhan Technical University of Surveying and Mapping.
1. INTRODUCTION
At present, digital orthoimages and orthoimage mosaics are playing an increasingly important role throughout the entire
geoinformation domain. Especially after the word of Digital Earth is put forward, digital orthoimages become the main
content of digital earth as well. Orthoimages offer an ideal combination of high information density, wide-area coverage,
economical and quick acquisition and production.
Because every orthoimage has relatively small area, the use of wide area is inconvenient. Certainly, several orthoimages
in which users are interested can be mosaiced into one big orthoimage, but it is inconvenient for users to access and
process.
The investigation and implementation of basic principles, data structure and efficient spatial index algorithm for Multiresolution Image Database are presented in this paper. Firstly the paper describes some basic concepts such as super
project, project, working area, network directory and the principle. The framework of image database is composed of
these concepts. Secondly the paper presents the data structure of image pyramid based on Multi-resolution. Thirdly the
paper presents a four-level index based on data structure of image pyramid. This index makes the whole database work
efficiently and it is also a core of the whole database. During the stages of developing the database, lots of orthoimages
from several provinces of China is used for testing. The conclusion of this study shows that this image database can
efficiently manage a mass of image data and that it can be used in geoinformation domain efficiently.
2.
THE BASIC CONCEPTS AND PRINCIPLES OF IMAGE DATABASE
2.1
The Basic Concepts
Before designing image database, some concepts are put forward as follows:
1. Image Workspace
In order to unify the orthoimages in different scale, conveniently intergeate with vector data and DEM data, and make
the image product standardized, we define the concept of Image Workspace.
Image Workspace is an image that belongs to a certain period and its size is equal to the map format. It is the smallest
unit in the image database. When the image is stored in the image database, image workspace data is formed after
organizing image file.
2. Map Format
Map Format is a set of orthoimages in a certain geographic area, including muti-resource and muti-temporal image data.
The size of Map Format generally is of the national standard, but it may be flexible sometimes.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B4. Amsterdam 2000.
1153
Mi Wang
3. Image Project
Image Project is a bridge that links different Map Format with the same scale. The Map Format with the same scale can
build one or several image projects. The data can be stored in any computer of local area network. Each image project
has independently spatial index and can make map format with the same scale seamlessly display and output in logic
4. Hyper Image Project
Hyper image project is also a bridge that links and manages all image projects. By hyper image project, we can
implement seamless display across scales and image projects.
5 Image Pyramid
In order to implement index of images with different scale and display images across scale, the data structure of image
pyramid is adopted. Image Pyramid can organize, store and manage muti-scale and multi-resource image data
efficiently. The lower level in the image pyramid has the higher resolution with a larger amount of image data. The level
number of image pyramid depends on practical needs.
6 Network Directory
The whole directory string which includes computer name, share directory name and image project directory is called
network directory. It exclusively identifies the location of image project and implement distributed image storage in
network environment.
2.2
The Basic Principles
Multi-resolution and Multi-resource Image Database is a spatial database system, where data object are orthoimages of
multi-resolution and multi-resource. On the basis of distributed file management system (RDBMS or Object-oriented
DBMS may be used. But considering the efficiency of image management, we prefer to use file management system.),
in order to implement image data distributed storage, the share directories in any computer are created in local area
network. The image projects with different scales will be stored in these share directories. The image pyramid is built.
The efficient four-level spatial index based on image pyramid can access image data conveniently and quickly. The
database system is seamless in geometric space and color space. It can manage mass image data and integerate with
other spatial database software on the same basis of spatial coordinate foundation. (Such as Geostar and GeoGrid
developed by Wuhan Technical University of Surveying and Mapping.)
3 THE DATA ORGANIZATION OF IMAGE DATABASE
The data organization of image database adopts the approach of object-oriented design. The whole image database
consists of many data access objects. The top object is hyper image project, which contains many sub-objects— image
project. Each image project contains many map formats, each of which contains many image workspaces.
Take account of the data structure, the whole system is a tree with hyper-project as root, project and workspace as
branchs and image as leaves. The data structure is shown as below in Figure 1:
HYPER IMAGE PROJECT
1:10000Image Project
Map Format 1
Image Workspace I
1:50000Image Project
Map Format 2
Image Workspace II
Image Workspace Ă
Map Format n
1: ? Image Project
Figure 1. The Data Structure of Image Database
1154
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B4. Amsterdam 2000.
Mi Wang
4.
THE MECHANISM OF SPATIAL INDEX BASED ON FOUR LEVELS MANAGEMENT
Image workspace, map format, image project and image scale are the foundations of four-level management in image
database. As shown in figure 1,they are:
1.Scale index based on the image window. According to this index, image scale to be used in image window is decided.
2.Image-project index based on the image window with the same scale. According to the scale that muti-scale index
finds, the same scale image projects across the image window are decided.
3.Map-format index based on the image window from the indexed projects above. According to the image projects that
the second index finds, map format across the image window are decided.
4.Image-workspace index based on the image type and image window. According to maps format that the third index
finds, the current workspace in each map format is decided.
Image
Window
Scale
index
Image-project
index
Map-format
index
Image-workspace
index
Figure 2 Diagram of Spatial Index
5. THE INTEGERATION WITH OTHER SPATIAL DATABASE
In order to update vector data in time, we may use image as the background. As image database has seamless character,
it is very convenient for wide area applications. Thus it is necessary to integrate vector data and image data.
On the other hand, if we want to make virtual reality more verisimilar in large area, we must use real texture image.
Through image database, the image with any area, any resolution can be provided as texture image.
As the spatial data in different spatial databases have the same mathematics foundation and spatial positoning
foundation. Thus it is possible to implement the integeration of spatial data.
GeoStar is software for GIS; GeoGrid is software for DEM. These software packages are developed by Wuhan
Technical University of Surveying and Mapping. During integeration we develop two DLLs GeoImageDBforGeoStar
and GeoImageDBforGeoStar using Visual C++ as the developing tool. Through DLL, GeoStar and GeoGrid can freely
access any range image data from image database. The sample of integeration is shown below:
Figure 3. The Integeration of Vector and Image Database
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B4. Amsterdam 2000.
1155
Mi Wang
Figure 4. The Integeration of DEM and Image Database
Figure 5. The Integeration of DEM, Liner graphics and Image Database
6.
CONCLUSIONS
During the research of image database, we did the experiments in Guangdong and Hainan Province of China. The basic
functions of image databases such as creating database, image query and image display are tested. The result proces to
be very good, which shows the value of image dtabase.
As image database is a new research direction, there are still many problems. So it is worthwhile doing research.
7.
ACKNOWLEDGMENT
The research described in this paper was funded by the Natural Outstanding Youth Science Foundation project of China
(No. 49525101)
8.
REFERENCES
Fang, T., Gong, J.Y.,Li, D. R., Main Key Techniques on Establishing Image Database, journal of WuHan Technical
University of Surveying and Mapping, Vol 22, No. 3,pp. 266-269
Schiewl J, Siebe E. Revision of Cartographical Database Using Digital Orthoimages, ISPRS Commission III
Symposium, 1994, 30(3/2).
1156
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B4. Amsterdam 2000.